Fig. 3: The results of BCNN with Laplacian-distributed, Gaussian-distributed and Bernoulli-distributed likelihood functions in simulated SPI with STL-10 dataset at 2×, 4×, 8× and 16× compression ratios.
From: Approximating the uncertainty of deep learning reconstruction predictions in single-pixel imaging
a A representative ground-truth image in the testing dataset, input images to the BCNN calculated from the LSQR-approximated inverse model matrix at the 2×, 4×, 8× and 16× compression ratios. b BCNN predictions with Laplacian-distributed, Gaussian-distributed and Bernoulli-distributed likelihood functions at the 2×, 4×, 8× and 16× compression ratios.
